Has anyone tried to to compare processing times between Standard Agents and Queue Processors? We are trying to kick off Queue Processor from a Job Scheduler. As the process is idempotent, the risk of locking errors while running it purely on Job Scheduler multi node setting is also low.
While comparing the numbers, we have noticed that the single thread queue processor is slower. Probably makes sense as we have additional step of queuing on Kafka and picking messages from there. The processing is at par with Job Scheduler when we run 3 threads. And there is no substantial improvement when we run 5 threads.
Has anyone else noticed similar behavior? We are well below the 20 partition limit, as we are running only two batch nodes in our UAT env.
***Edited by Moderator: Pooja Gadige to add capability tag***